https://netmaker.org logo
Powered by
Hey gang! Having some issues with joining my netwo...
# client
f
Hey gang! Having some issues with joining my network on a linux client. When joining I'm getting some IPv6 (I guess?) related issues... How can I solve this?
Copy code
[netclient] 2022-09-07 20:40:25 error installing:  error creating node 500 Internal Server Error {"Code":500,"Message":"address fe80::6842:456a:b567:fc18:51821: too many colons in address"}
j
what is your join command?
and what does your network look like? dual stack with udp hole punching off?
f
No, IPv6 is turned off 🤔
join command is the usual "netclient join -t "
Hole punch is enabled
j
is that the MAC of the machine?
nvm
i have no clue where that could possibly be coming from
f
😅
j
have you tried leaving and re-joining? make sure the config is wiped under /etc/netclient/config when you re-join
f
Yup, same issue.
j
is this happening on all linux clients?
and im guessing this is 0.15.1
f
The node seem to be created and shows in the GUI, but it isn't working as nothing is being stored in /etc/netclient/config
Haven't tried on another linux client yet.
Yes, 0.15.1
j
server was upgraded too?
f
Yup!
Just tried creating a new network, and could join that
The network that I can't join has been through a few version upgrades...maybe something has happened between the upgrades.
j
did you turn ipv6 on/off?
f
off
as my troublesome network
j
man...it's an ipv6 address with the default port stuck onto the end of it...I have no idea where/how that could possibly happen
@bored-island-21407 @most-rain-82770 for visibility
f
haha, yeah that was odd.
Can I adjust this manually, or should I just scrap the network and recreate?
j
did you delete the node from the GUI before attempting to re-join?
im really not sure, I'm not able to re-create the issue. It would be good if you're able to gather some more info on it, and maybe try joining from another machine to see if you get the same issue
i have to go for now, will be back soon
f
Yes, deleted the node before join
Alright, I can test to join from another machine
Huh... Tried some netclient pull on first one of my unraid machines (based on slackware) and got the same error where the port is added to the ipv6 address. Then I tested on another linux machine (ubuntu), and there the pull went fine 🤔
b
any logs from server?
f
Let me check
in the netmaker logs I see a bunch of
Copy code
[netmaker] 2022-09-07 19:34:38 error getting peer update for node d0db1e6e-98a7-47a5-bd00-4fe1ccc59c58 address fe80::6842:456a:b567:fc18:51823: too many colons in address
[netmaker] 2022-09-07 19:34:38 error getting peer update for node 020d8490-e166-4cdd-bbe8-e070e34e722b address fe80::6842:456a:b567:fc18:51823: too many colons in address
[netmaker] 2022-09-07 19:34:38 error getting peer update for node 399a7ca5-51cb-4a6d-8a1c-6890073c44af address fe80::6842:456a:b567:fc18:51823: too many colons in address
I think those three nodes are the only ones with that error message
b
sqlite3 /var/lib/docker/volumes/root_sqldata/_data/netmaker.db "select * from nodes where key=d0db1e6e-98a7-47a5-bd00-4fe1ccc59c58;"
f
Copy code
Error: unrecognized token: "98a7"
ah, adding ' around the key
So that's the node running pop-os
And the other two nodes are a Windows laptop, and lastly one of my Unraid machines.
But where do this ipv6 address with the attached port number come from? 😄
b
i am not sure
f
Do you need the query output?
b
not in address or address6 of the node
only if that ip6 address: port combo is in there somewhere
f
It's not there, I can't see any ipv6 address in there at all
b
good to know
you got the error with a join and with a pull, correct?
f
Correct
Hmm, I found the ipv6 address now at least... apparently it's a windows node (which currently is offline as that machine is turned off)
b
is that the local address of the windows machine?
f
Yup, "localaddress":"fe80::6842:456a:b567:fc18"
But that node should be using the default port, not 51823 that can be seen in the netmaker logs
b
wonder if it thinks that windows machine has the the same public ip as another node and thererfore should use the local address as the peer endpoint
f
ahhh, yes. The three nodes with that error is on the same network, but having different ports.
b
If you edit the local address of the windows node in the ui, does the problem go away
f
Hmm, is it reasonable to think that bringing the machine online solved the problem? Because now the pop-os node could join again
and I didn't edit the local address for the windows node
b
Yes, when that machine came online, it should have updated its local address.
f
Ah
b
Does it still show the ipv6 address for local address for that node in ui?
f
Yes 🙂
Or no I mean
😄
It's an ipv4 address now
b
ok, that makes sense.... what doesn't make sense is why the local address got set like that in the first place.
f
I see that the client is still running 0.14.6... but that's maybe unrelated
on the windows node that is
b
so I know where the error comes from
just baffled on how the local address got set to that
f
Now you got me curious 😄
b
if you pass an ipv6 address:port to net.ResolveUDPAddr you have to encase the ipv6 address in brackets [ ]
f
ah 🙂
b
net.ResolveUDPAddr("udp", "[fe80::6c85:a45f:27a1:2fff]:51820") works net.ResolveUDPAddr("udp", "192.168.0.1:51820") works net.ResolveUDPAddr("udp", "fe80::6c85:a45f:27a1:2fff:51820") thows error --- too many colons in address
still baffled how local address got set to an ipv6 address. we make a connection to an ipv4 address and check the local end of that connection
f
Yes I have no clue either. I'm fairly certain that I've never fiddled with the local address for a node manually 🙂
j
wow, that's a weird one
we really need to refactor the way we do local routing
glad you found it though. Thanks @bored-island-21407 for helping on that
q
Do you know any workaround for this issue? I have the same issue with public ipv6
2 Views
PreviousNext
Powered by